Molecular diagnostic tests with application to clinical diagnostics involve studies in infectious diseases, inherited diseases, oncology, predisposition to disease, or the description of polymorphisms linked to disease states. General considerations in the design of evaluation of diagnostic test trials and statistical principles for reporting the results are discussed. A brief overview of the general statistical considerations related to the intent of use, test development versus validation, different types of biases, and issues with missing data are provided. Furthermore, issues related to commonly used but not necessarily correct methods to characterize the performance in the presence and absence of a clinical reference standard are discussed. These issues are broadly applicable to any molecular diagnostic test with a dichotomous result. This overview may help the clinical molecular diagnostic community to evaluate tests that provide a dichotomous result.
Published by Elsevier Inc.